A Long, Long Time Ago
It’s 2013 in beautiful Knoxville, Tennessee and I’ve just been hired by a healthcare software company to build up a UX design team. The focus? Physician-facing and care team facing applications and workflows. The kind of context where deep knowledge of how clinical users actually think and work isn’t a nice-to-have, it’s a requirement.
Except that’s not quite what happened. Over the following months, through a series of smaller decisions and quieter conversations, the picture changed. The team never materialized and the role drifted. What they actually needed, it turned out, was someone to evaluate vendors and manage design relationships. An art director, something I had done before, pretty extensively. And yet…
The contextual knowledge that healthcare UX requires, the kind you build through research and iteration and repeated proximity to actual users? It got treated as a secondary consideration. Engineering was the center of gravity. Design just orbited it. I trained my replacement and left exactly one year after I started.
Two years later
At Siemens Healthineers, I’m sitting across from an engineering manager who has politely but clearly already decided this meeting is a formality.
We’re talking about PET scanner workflow software. The screens that a nuclear medicine technician uses to dial in radiation sensitivity, read radiological event data and confirm what’s been captured and what’s still pending. Heady stuff.
The screens are all in black and white. Dense gray scale without much visual hierarchy. No color cues and no summary of present state. Just numbers and toggles that the technicians had been memorizing for years, because the alternative was slower.
I had some questions, and with tech constraints onboarded, I soon had upgrade proposals: better resolution, colors to encode and depict radiation sensitivity ranges, a progress summary surfaced where it was actually needed and changes that would reduce the cognitive load in workflows where errors had real consequences.
The engineering manager looked at the mockups, and then at me, then back again at the mockups. “We’ll think about it,” he said. That quarter, I ate a lot of catered lunches at my own brown bag events talking to engineers, clinicians and physicians as well as product managers, sales and marketing.
Doing Reps – UX Adoption
Here’s what UX adoption actually looked like in that era, from the inside.
Perhaps you were in a new practice trying to earn a seat at a table that had been set without you. Engineers and executives didn’t dismiss UX because they had evaluated it and found it wanting, but because it looked decorative. Soft and touchy-feely. The frou-frou department, one executive at another company once called us. (I’m not making that up.)
The resistance lived at the top and in the middle and the people with institutional authority were the skeptics. Which meant that one of the most promising paths forward was bottom-up proof.
Meaning what? Patient, persistent and high-quality work that kept raising the bar. Usability testing session videos where engineers sat and watched real users struggle with interfaces they had built. HCI fundamentals explained in terms of sensory memory and visual processing, in the language of people who thought in systems. Brown bag lunches and hallway conversations. Prototypes that were better than anything anyone had seen before from a design function.
One full quarter of pursuing this at Siemens Healthineers and then…Then a YES.
Two companies, two shapes
They both circled the same problem: UX climbing uphill against institutional skepticism, converting the powerful through evidence they couldn’t easily dismiss.
There’s a documented failure mode worth naming here, one which Debbie Levitt wrote about in 2022 in UX Magazine: UX practitioners who responded to that skepticism by democratizing their work, running design sprints and workshops that invited everyone to “do UX,” inadvertently taught organizations that UX was something anyone could do. The tactics designed to build buy-in ended up hollowing out the credibility they were meant to establish.
The ones who navigated it well held the line on expertise while still opening the door. Proof positive rather than participatory theater.
AI arrived differently
It didn’t send a calendar invite. It didn’t ask for a pilot program. The budget showed up first, and then the mandate. Then the all-hands where leadership explained that we were going to be an AI-forward organization, effective immediately, and wasn’t that exciting?
The resistance to this new influence flipped from technical and leadership to creatives and engineers in the trenches.
We’re talking Design and Development, the people closest to the craft. Everyday users who had been handed tools they didn’t ask for and told to use them. These are the skeptics now. Not the executives, and not much engineering leadership. The people with institutional authority are already sold while the unconvinced are the practitioners.
Same adoption problem, with an opposing polarity. So what does that reticence actually look like up close? Some of it is displacement anxiety, which is real and reasonable. Some of it is quality concern, also real. Some of it is ethical unease about training data, authorship, environmental cost. Some of it is something harder to name: the feeling that something is being done TO you rather than WITH you. That the conversation started without you and ended before you arrived. That last one sucks, doesn’t it?
Resistance Vectors
Here’s what I keep returning to… That the tactics that worked in 2013 were designed for a specific direction of resistance.
Brown bag lunches work when the VP needs convincing. They don’t work when the VP is already sold and your creative director is the holdout. You can’t mandate your way to bottom-up trust. And you can’t run a workshop that teaches skeptical practitioners that “AI is good, actually” without recreating the exact failure mode Levitt described. You’d be running AI evangelism theater, basically going through motions that signal organizational enthusiasm while building neither genuine capability nor genuine confidence in the people you most need to reach.
So what does the reverse playbook look like? I don’t think we have a settled answer yet, but some hypotheses are worth examining.
Proof by invitation, not assignment. Let skeptics choose their own first use cases rather than handing them one. The brown bag that worked at Siemens wasn’t mandatory. It was catered, which is different. People showed up because they wanted to, and left having seen something they hadn’t expected.
Honest failure visibility. Show where AI breaks, not just where it succeeds. Calibrated trust is more durable than converted enthusiasm. The usability testing videos that moved engineers at Siemens weren’t highlight reels. They were uncomfortable to watch, and that was a big part of the point.
Craft preservation framing over craft replacement framing. What does AI protect practitioners from, rather than what does it take? That reframe isn’t just spin, and more often than not it’s a more accurate description of what’s happening in the workflows where AI is genuinely working well.
Peer signal over executive signal. Who in the skeptic community is actually using these tools well, and are they visible? In 2015, the practitioner who moved me forward with engineering management wasn’t the VP. It was a senior engineer who had sat through a usability session and couldn’t stop talking about it afterward. Credibility traveled laterally before it traveled up.
None of these are prescriptions, and obviously the playbook for these tactics is still being written.
Next Generation Adoption Attitudes
One more layer of this problem is being built right now, and it’s further out than many folks are looking.
On a recent road trip, my teenager told me about AI at school. Compulsory exposure. Tools introduced without much context for why. Classmates who had concluded, from their first extended contact with AI, that it was something to be skeptical of or dismissed. Not because they’d evaluated it carefully, but because their introduction felt like something being done to them.
Sound familiar? The adoption problem isn’t just in current workplaces. It’s being reproduced downstream in a generation whose first impressions are forming right now. That’s a longer conversation and it deserves its own article. But it’s worth naming here: we’re not just behind on the current adoption curve. We may be behind on the next one too.
Wrapping Up
In 2013 the question was: how do you convince the people with power that this practice has value? Well, we figured that out. Imperfectly and slowly, with more catered lunches than anyone should have to sit through. But we figured it out.
The question now is different. How do you build genuine trust with the people doing the work, when the people with power already believe?
I don’t think we’ve found the answer yet, but I’d love to talk with the bright minds working on it.
—–
#AIAdoption #AIAdoptionDebt #DesignOps #UX #Leadership #ProductManagement #EngineeringLeadership #FutureOfWork #NextGenerationTalent